2020–2025; closer lines indicate more consistent extraction
Comparing total injuries at each severity level across both test runs to check for consistency. Divergence between the two lines suggests the extraction is producing different results for the same articles.
2020–2025
2020–2025
2020–2025
2020–2025; top 20 states by combined article count
2020–2025
Each point = one value; line = median. Higher is more consistent.
Distribution of Jaccard % agreement scores across all values within each data category. Each point is one factor, state, or grade. The vertical line shows the median for that category.
Climbers tagged as a fatal injury in one run but not the other. Each row represents a case where the two runs disagreed on whether someone died.
Analysis by Nate Downer
Social Risk Factors
2020–2025